One-Day-Ahead Load Forecasting using nonlinear Kalman filtering algorithms

In this paper, we consider the problem of 24-hour ahead short-term load forecasting; the formulation is based on the nonlinear Kalman filtering. Our formulation takes into account weather conditions as well as previous trends. Effects of weather as well as prior consumptions are nonlinear functions; hence our choice. We compare our proposed method with the standard Kalman filtering approach and with the state-of-the-art echo state network. Experiments are carried out on the well known REDD dataset. We show that our proposed nonlinear Kalman filtering algorithm outperforms all prior techniques. © 2017 The Authors. Published by Elsevier Ltd. Peer-review under responsibility of the organizing committee of SP-CRTPNFE 2016.